Running tests is nice as long you hold the experimenters accountable to actually measure the effect.

For example, educational programs sound as a rational idea until you look at older studies that looked at the effect of increasing the punishment for crime. As I recall there were no effect for most form of violence, including sexual violence, with the researchers conclusion that much crime don't follow rational planning. If the criminal don't do a risk vs reward analysis before the crime, then increasing the punishment won't have an effect on the thought processes. Similar education and work sponsored events is unlikely to help if the action is irrational and involve a lot of self-delusions.

I can see how mores lights and banning hard alcohol can work. People have a emotional reaction to light, and drugs/alcohol operate on the part of the brain that handles social interaction and inhibitions.